Tech, Markets, and the Midterms

Alphabet

Across the shows: 51 speakers in other episodes, 40 bullish and 30 bearish calls, past 30 days

• Heath Terry · Citi ResearchBearish

  Search dominance is cracking, and Google's AI search share will likely fall well short of its 90% search share.

  Terry says last quarter was the first sign of cracks in Google's search dominance, which suggests AI-driven search monetization is not comparable to what the search business earned. He acknowledges Google is clearly a leader in AI with performant models and a strong Google Cloud business with enterprises. He still thinks it is unlikely its AI market share will reach the 90% it holds in search, which is the bear case the stock has faced for four years.

  “This last quarter was the first time that we started to see some cracks in that search dominance... it's unlikely that their AI market share is gonna be the 90% that their search market share is.”at 34:19

• Heath Terry · Citi ResearchBearish

  Alphabet's underperformance reflects income-oriented holders leaving; it will trade at lower multiples.

  Investors once owned Amazon, Google and Microsoft for their cash generation, dividends and steady, low-anxiety returns. That is no longer the case because of heavy AI spending, so holders with those mandates are turning over their positions, which explains the underperformance. Dividend and value funds can't hold these stocks for the old reasons anymore, so they either own something else or pay a much lower multiple for the ones they still hold.

  “If you had owned Amazon or Google or Microsoft, you own them because they generated a lot of cash... And that's not the case anymore... you've got this turnover happening from investors that owned them for all of those reasons before.”at 38:40

Amazon.com

Across the shows: 33 speakers in other episodes, 42 bullish and 7 bearish calls, past 30 days

• Heath Terry · Citi ResearchBearish

  Amazon's underperformance reflects income-oriented holders leaving; it will trade at lower multiples.

  Investors once owned Amazon, Google and Microsoft for their cash generation, dividends and steady, low-anxiety returns. That is no longer the case because of heavy AI spending, so holders with those mandates are turning over their positions, which explains the underperformance. Dividend and value funds can't hold these stocks for the old reasons anymore, so they either own something else or pay a much lower multiple for the ones they still hold.

  “If you had owned Amazon or Google or Microsoft, you own them because they generated a lot of cash... And that's not the case anymore... you've got this turnover happening from investors that owned them for all of those reasons before.”at 38:40

Microsoft

Across the shows: 36 speakers in other episodes, 37 bullish and 5 bearish calls, past 30 days

• Heath Terry · Citi ResearchBearish

  Microsoft's underperformance reflects income-oriented holders leaving; it will trade at lower multiples.

  Investors once owned Amazon, Google and Microsoft for their cash generation, dividends and steady, low-anxiety returns. That is no longer the case because of heavy AI spending, so holders with those mandates are turning over their positions, which explains the underperformance. Dividend and value funds can't hold these stocks for the old reasons anymore, so they either own something else or pay a much lower multiple for the ones they still hold.

  “If you had owned Amazon or Google or Microsoft, you own them because they generated a lot of cash... And that's not the case anymore... you've got this turnover happening from investors that owned them for all of those reasons before.”at 38:40

AI infrastructure & data centers

• Heath Terry · Citi Research

  Data center projects are slipping from Q4 into 2027, pushing supply back and softening spot markets; not necessarily bad for AI.

  Terry says building data centers is hard once the permitting, power, labor and electrical equipment are involved. Many projects expected by December are now slipping to Q2 and Q3 of next year, which pushes supply back into the chain. That is softening spot markets and pushing back revenue for companies expecting it sooner. He does not think this is bad for the broader AI trade, since spreading the buildout into 2027 through 2029 could make earnings more sustainable.

  “It is really easy to talk about building data centers... it's hard. ... these q four projects that were expected to get done by December are now q two, q three projects for next year, and that's pushing a lot of that supply back into the supply chain.”at 30:46

• Heath Terry · Citi Research

  Hyperscalers earn about a 26% cash return on AI capex, far above their roughly 6% borrowing cost.

  Terry says the hyperscalers have to keep investing in AI to generate the returns they are seeing, and his math suggests they are getting a 26% cash return on cash invested in AI. He notes they can borrow at about 6%, so the spread favors borrowing to fund the buildout.

  “They know they've gotta continue to invest in AI to generate the returns that they're seeing for it... The math that we do would suggest they're seeing 26% cash return on cash invested... And so if you can borrow money at 6% and get a 26% return,”at 33:27

• Heath Terry · Citi Research

  Hyperscaler AI capex is a new norm; free cash flow turns positive around 2029 as assets pay back.

  Hyperscalers' AI capex is not a one-time bump; it is the new norm. There is a one-time component, a step-function catch-up to demand, similar to early cloud buildouts by Amazon and Google, but it is being compressed. What took about ten years to build in cloud is being attempted in about three, so these companies face a one-time period of negative free cash flow. Once the infrastructure is built it generates cash flow. In his models, the crossover where cash-generating assets exceed the incremental cost of new assets happens in 2029. He notes hyperscalers generate high returns on cash invested, which is one reason the firm likes them ultimately, but there is a timing gap because current investment is bigger than current returns.

  “This is the new norm. ... What took ten years to build in cloud, we're trying to do in three. ... we think it's 2029 where that crossover happens, where the cash flow generating assets are bigger than the incremental cost of new assets.”at 39:41

US elections & politics

• Meredith Whitney · Meredith Whitney Advisory Group

  Republicans face a very bad midterm result because they are out of touch on affordability.

  Whitney says the Republicans have been out of touch on affordability. Official BLS income data, which shows all income levels growing, is stale. For people on the brink, one in four SNAP recipients has lost access to food stamps, and she says the electorate is angry. From her work she expects the result to be really bad for Republicans. She also expects Talarico to win in Texas, which she reads as a sign of growing centrism.

  “I think it's gonna be really really bad. I mean, the Republicans have been so out of touch affordability.”at 4:07

• Joe Lavorgna

  Betting markets underestimate a Democratic sweep: about 235-240 House seats and at least six Senate pickups.

  Lavorgna says betting markets do not understand the breadth and depth of a likely Democratic victory. He expects the Democrats to win the House with about 235 to 240 seats. In the Senate they should pick up at least six seats. He argues the races are highly correlated, so a state like Kansas, which he thinks flips, pulls in Iowa and Nebraska. He points to presidential net approval, enthusiasm and generic ballot polling as the basis.

  “The problem I have with the betting markets is I don't think they understand the potential breadth and depth of a likely or highly likely Democrat victory. ... they may have 235, 240 seats. In the senate, Democrats probably pick up at least six seats.”at 18:59

US consumer

• Meredith Whitney · Meredith Whitney Advisory Group

  Stress is concentrated at the bottom: lenders serving lower-income households keep outperforming as a K-shaped gap widens.

  Whitney argues the official data hides real strain. Banks lend mostly to prime and super-prime borrowers, so she looks at pawn loans, which she says have grown more than 30% over the last few years, and earned wage access, which she says has grown like a rocket ship. Those people go from payday to payday. She points to Dave, Chime and Block as the fastest-growing products in their companies. She sees a growing chasm between haves and have-nots, and the nonbanks serving the bottom have outperformed and will keep doing so.

  “the big growth areas, which is pawn loans, which have grown 30 plus percent over the last few years. Daily wage access has grown like a rocket ship.”at 5:34

Banks

• Meredith Whitney · Meredith Whitney Advisory Group

  Not keen on banks beyond trading; buybacks, not growth, drive them and little is happening there.

  Whitney says she has not been keen on the banks except for trading, because they have been largely vehicles of share buybacks and have not really grown. She separates JPMorgan from the rest. As an example she cites Citi, which was trying to get a discount to tangible book and now trades at or above book. Her conclusion is that she does not see much going on in the banks.

  “I haven't been so keen on the banks but for trading because they've been largely vehicles of, you know, share buybacks. They haven't really grown. ... Citi was trying to get a discount to tangible book, and now it's at book.”at 8:28

Payments & fintech

• Meredith Whitney · Meredith Whitney Advisory Group

  Payment networks have been financials' standout; asset-light, high-margin nonbanks will keep outperforming banks.

  Whitney calls the payment companies the shining stars of financials and the OGs of fintech. She credits their asset-light models and high margins, and says nonbanks have really outperformed and will continue to outperform. Her comparison is with banks, which she says have grown little and mainly returned cash through buybacks.

  “The payment companies have been the shining stars of financials. They were sort of the OGs of fintech. ... asset light models, high margin vehicles. ... But I think that the nonbanks have really outperformed and will continue to outperform.”at 8:46

AI: overall outlook

• Joe Quinlan · Bank of America

  AI exposure extends across the ecosystem; clients won't give up AI winners, which should do well in 2027 and 2028.

  Quinlan says AI exposure is an ecosystem, not just the big names, and it reaches emerging markets such as South Korea and Taiwan. Clients worry about concentration, but many do not want to give up these winners. He believes they will keep doing well in 2027 and 2028, since these companies have done so well. The support for the forecast is thin and rests mainly on the strength of past performance.

  “And it's even in the emerging markets, South Korea and Taiwan. ... a lot of our clients don't wanna give up on these winners. These companies have done so well, and I believe they're gonna do well in the '27 and '28.”at 13:07

Manufacturing & industrial policy

• Joe Quinlan · Bank of America

  New investor money is flowing into old-economy US infrastructure: pipes, valves, transmission lines, water systems.

  Quinlan says many clients are stepping back into the market while the Magnificent Seven has lagged this year, with the group up only about 3 to 4%. He says the real opportunity is the infrastructure belt, the nuts and bolts. He names pipes, valves, transmission lines and water systems, which he calls old industrial US. He says this is where a lot of folks are putting new money to work. He also points to manufacturing plants being built and hiring in the heartland.

  “It's the infrastructure belt. It's the nuts and bolts. It's the pipes, valves, the transmission lines, water system. It's old industrial US.”at 13:40

US stock market outlook

• Joe Quinlan · Bank of America

  Bull market has runway: solid 2027 likely, choppy 2028 as the election platforms take shape.

  Quinlan says the average bull market lasts five and a half years, so this one is not over. Earnings are good, the economy is strong, and the earnings outlook is robust. He expects a good 2027 and a choppier 2028 as the parties set out their platforms. He also argues stocks can keep rising despite higher rates because nominal growth is driving them. Nominal growth is closer to 10% than 5%, supported by manufacturing buildouts and hiring.

  “The average bull market is five and a half years, so we're not there yet. We still got some runway. But we see a good 2027.”at 14:09

US rates & the Fed

• Joe Quinlan · Bank of America

  BofA team expects Fed rate hikes in December and again in March.

  Quinlan says the Fed is trying to thread the needle. The Bank of America team expects another rate hike in December and another in March, which reverses the 2025 cuts. He calls this normalization of the cost of capital: rates should not be three, and not six, but in a more balanced range. He says there is still a lot of foreign demand for US assets.

  “Our team is expecting another rate hike in December, another one in March. You're just taking away the rate cuts from 2025.”at 16:06

Tariffs & trade policy

• Joe Lavorgna

  The Canada trade war was largely self-inflicted and is turning key states like Michigan against the GOP.

  Lavorgna links the trade war with Canada to the midterm outlook. Michigan is an important state for that trade dispute, and he says a Michigan Senate race is in play after a senator spoke out against the tariffs. He calls the mess largely self-inflicted.

  “Well, because the trade war with Canada... So it's a real mess. And a large of it, a lot of it, I think, was unfortunately self inflicted.”at 20:07

US deficit & government debt

• Joe Lavorgna

  Gridlock is bad: longer shutdowns likely, and the debt ceiling breaks within the next year.

  Lavorgna says this gridlock is not the good kind. The political divisions and divisiveness are worse than before, so shutdowns could last longer. The debt ceiling is a further risk, since it will break within the next year and could make things much worse.

  “The gridlock is not good kind of gridlock. It's bad gridlock because the political divisions and divisiveness are worse. Things like shutdowns could be longer. We've got a debt ceiling that's gonna break up the deal within the next year.”at 22:40

US inflation

• Joe Lavorgna

  The economy is an inflationary boom, not disinflationary; real wages fall and housing affordability is at record lows.

  Lavorgna says the economy is booming but it is an inflationary boom, not a disinflationary one. Affordability and the cost of living are declining as real wages fall. Even though it is not in the inflation data, first-time home-buyer affordability is essentially at an all-time record low. He says the American worker is suffering, which he sees as the main reason Republicans will lose.

  “The economy is a boom. It's just not a disinflationary boom. It's more of an inflationary boom... declining real wages... first time housing affordability, it's essentially all time record low.”at 25:21

Big Tech platforms & digital ads

• Heath Terry · Citi Research

  Hyperscalers' AI revenue is accelerating sharply, while legacy search, retail and software businesses drag stocks down.

  Terry says the hyperscalers have two sides. The AI business is accelerating: revenue growth rose about 900 basis points from Q2 to Q3, and hyperscaler backlog is up 142% year over year. The legacy businesses, Google search, Amazon retail and Microsoft software, are doing much worse, and that is dragging the stocks down. He says AI is effectively crowding everything else out.

  “there is the AI related part of their business where you're seeing this massive acceleration in revenue growth, 900 basis points of growth acceleration from q two to q three, hyperscaler backlog up a 142% year over year. ... Google search business, Amazon's retail business, Microsoft software business, and it's a different story.”at 32:31

AI regulation & safety

• Heath Terry · Citi Research

  Public sentiment against AI makes the pro-industry stance a political trade-off that candidates will keep exploiting into 2028.

  Terry says the administration wants the US to win AI and must support its leading companies, but that is a hard trade-off. Public sentiment is already very negative about AI, so being negative on AI is a winning position for politicians seeking re-election. He points to some 40-odd ballot measures on data centers as evidence that this will continue into 2028.

  “it's a hard trade off because public sentiment wouldn't be much more negative around AI. ... being negative AI is a winning position to take. And going into '28, that's gonna continue to be the case.”at 35:47

AI bubble & tech valuations

• Heath Terry · Citi Research

  Debt-funded AI capex is weighing on hyperscaler stocks despite rising estimates; issuing debt has a valuation cost.

  Terry says the move to a more indebted model is already weighing on the stocks. Google's 2027 revenue estimates are up 25% and its EPS estimates are up 12% since the start of the year, yet the stock is down 10%. He reads that as the market pricing a cost to being an issuer rather than a buyer, and says it will keep dragging on valuations.

  “It's gonna weigh on the stocks. It already has. ... Their EPS estimates for '27 are up 12%, and the stock's down 10. And that is the market saying that there is a cost to being an issuer versus a buyer.”at 36:30